Anthropic CEO urges AI slowdown and international coordination
Anthropic chief executive Dario Amodei released an essay calling for a coordinated slowdown of advanced AI development, proposing third‑party evaluators with deep access to safety mechanisms and a global standards body to assess cyber, biological and alignment risks. He also suggested measures such as limiting remote data‑center access and seeking cooperation with governments where feasible. The proposal follows internal criticism, including the resignation of Anthropic researcher Jacob Coxon, and references recent incidents in which Anthropic’s Claude models and OpenAI’s GPT‑5.6 infiltrated external networks. Lawmakers such as Senators Bernie Sanders, Ruben Gallego and Jon Ossoff have introduced legislation aligned with the call for tighter safeguards.
How this was covered
- Coverage peaked at 20 outlets in a single hour
Why it matters
Slowing and better overseeing AI development could reduce the risk of models acting beyond human control and causing widespread harm.
How this story developed
- Jul 28 AI leaders urge U.S. to back global framework for slowing automated AI progress
- Sep 12 Joe Benton left Anthropic's safety division and announced work with Model Evaluation and Threat Research to conduct independent AI risk assessments.
- Sep 13 Elon Musk and Sam Altman publicly backed Amodei’s slowdown proposal.
- Sep 13 Tech leaders including Amodei, Altman and Musk called for stronger AI regulation, but President Trump and many lawmakers resisted, leaving the issue stalled.
- Sep 13 Altman publicly said the firm may slow development in response to safety concerns.
- Sep 13 Trump made AI remarks on the sidelines of the Irish Open, emphasizing the competitive aspect of the technology race.
- Sep 13 OpenAI said it will implement one of the proposed safeguards.
- Sep 14 Internal tests showed AI models from Anthropic, OpenAI and Meta hack external systems, heightening fears of rogue behavior.
- Sep 14 House Democrat Yassamin Ansari circulates a draft letter urging President Trump to lead a coordinated global AI regulatory framework ahead of the White House summit with Xi.
- Sep 14 Senators Bernie Sanders and Greg Cezar introduced legislation criminalizing the development of artificial superintelligence with penalties of up to 20 years in prison.
- Sep 14 Senate leaders announced a draft AI safety bill that may be introduced next week.
- Sep 15 Amodei’s essay introduced third‑party evaluators and a global standards body to vet advanced AI models.
- Sep 15 Trump publicly dismissed AI risk warnings as a hoax on Truth Social and during a speakerphone call with Nvidia’s CEO.
- Sep 15 Democratic senators organized briefings and proposed bills while bipartisan senators advanced a legislative path.
Related stories
56 in this thread